The abiotic and biotic drivers of rapid diversification in Andean bellflowers (Campanulaceae)
The Harvard community has made this article openly available. Please share how this access benefits you. Your story matters
Citation Lagomarsino, Laura P., Fabien L. Condamine, Alexandre Antonelli, Andreas Mulch, and Charles C. Davis. 2016. “The abiotic and biotic drivers of rapid diversification in Andean bellflowers (Campanulaceae).” The New Phytologist 210 (4): 1430-1442. doi:10.1111/nph.13920. http://dx.doi.org/10.1111/nph.13920.
Published Version doi:10.1111/nph.13920
Citable link http://nrs.harvard.edu/urn-3:HUL.InstRepos:27822355
Terms of Use This article was downloaded from Harvard University’s DASH repository, and is made available under the terms and conditions applicable to Other Posted Material, as set forth at http:// nrs.harvard.edu/urn-3:HUL.InstRepos:dash.current.terms-of- use#LAA Research
The abiotic and biotic drivers of rapid diversification in Andean bellflowers (Campanulaceae)
Laura P. Lagomarsino1, Fabien L. Condamine2, Alexandre Antonelli2,3, Andreas Mulch4,5 and Charles C. Davis1 1Department of Organismic and Evolutionary Biology, Harvard University Herbaria, Harvard University, Cambridge, MA 02138, USA; 2Department of Biological and Environmental Sciences, University of Gothenburg, Goteborg€ SE 405 30, Sweden; 3Gothenburg Botanical Garden, Goteborg€ SE 413 19, Sweden; 4Senckenberg Biodiversity and Climate Research Centre (BiK-F), Senckenberg, Frankfurt/Main 60325, Germany; 5Institute for Geosciences, Goethe University Frankfurt, Frankfurt/Main 60438, Germany
Summary Author for correspondence: The tropical Andes of South America, the world’s richest biodiversity hotspot, are home to Laura P. Lagomarsino many rapid radiations. While geological, climatic, and ecological processes collectively explain Tel: +1 314 577 0285 such radiations, their relative contributions are seldom examined within a single clade. Email: [email protected] We explore the contribution of these factors by applying a series of diversification models Received: 27 October 2015 that incorporate mountain building, climate change, and trait evolution to the first dated phy- Accepted: 26 January 2016 logeny of Andean bellflowers (Campanulaceae: Lobelioideae). Our framework is novel for its direct incorporation of geological data on Andean uplift into a macroevolutionary model. New Phytologist (2016) 210: 1430–1442 We show that speciation and extinction are differentially influenced by abiotic factors: spe- doi: 10.1111/nph.13920 ciation rates rose concurrently with Andean elevation, while extinction rates decreased during global cooling. Pollination syndrome and fruit type, both biotic traits known to facilitate mutu- Key words: Andes, biodiversity hotspot, alisms, played an additional role in driving diversification. These abiotic and biotic factors climate change, diversification, Lobelioideae, resulted in one of the fastest radiations reported to date: the centropogonids, whose 550 Neotropics, pollination syndromes, rapid species arose in the last 5 million yr. radiation. Our study represents a significant advance in our understanding of plant evolution in Andean cloud forests. It further highlights the power of combining phylogenetic and Earth science models to explore the interplay of geology, climate, and ecology in generating the world’s biodiversity.
> 60% of the current elevation of the central Andes was attained Introduction within the last 10 million yr (Myr) (Gregory-Wodzicki, 2000; Species-rich rapid radiations are a conspicuous ecological and Garzione et al., 2006, 2008, 2014). Although the onset of evolutionary phenomenon in the Tree of Life. Clades that have Andean orogeny dates to the Paleocene and Eocene (66– undergone such diversification are often documented in insular 33.9 Myr), including in regions as far north as the modern East- environments, including islands (Baldwin & Sanderson, 1998; ern Cordillera of Colombia (Parra et al., 2009), exceptionally Givnish et al., 2009; Lapoint et al., 2014), lakes (Wagner et al., rapid surface uplift occurred during the late Miocene (c.10– 2012), and mountains (McGuire et al., 2007; Hoorn et al., 2013; 6 Myr) and early Pliocene (c. 4.5 Myr) (Garzione et al., 2008; Hughes & Atchison, 2015; Merckx et al., 2015). Although they Hoorn et al., 2010; Mulch et al., 2010). Such mountain building represent just one-eighth of terrestrial land surface, mountains is thought to promote diversification in a variety of ways, includ- are home to one-third of all species (Antonelli, 2015) and a large ing by increasing physiographic heterogeneity, affecting local and number of species-rich radiations (Hughes & Atchison, 2015; regional climate, facilitating the immigration of preadapted Schwery et al., 2015), including some of the fastest diversification species, and creating opportunities for extensive, parallel geo- rates reported to date (Madrin~ an et al., 2013). Of particular graphic speciation and adaptive radiation in island- and importance are the tropical Andes, which stretch from Venezuela archipelago-like venues (Hoorn et al., 2013; Givnish et al., 2014; to northern Argentina along the western coast of South America. Mulch, 2016). These incredibly species-rich mountains (Barthlott et al., 2005; The role of the rise of the Andes in the origination of biodiver- Kreft & Jetz, 2007) are home to c. 15% of all flowering plant sity has been implicated in clades as diverse as birds (McGuire species, half of which are endemic to the region (Myers et al., et al., 2014), butterflies (Elias et al., 2009), and angiosperms 2000). The extent of this biodiversity is especially striking consid- (Hughes & Eastwood, 2006; Madrin~ an et al., 2013). Within ering the recency of mountain uplift: despite debate over the pre- angiosperms, previous work has emphasized that the timing cise timing and rates of uplift (Sempere et al., 2006; Ehlers & and geotemporal trajectories of diversification have been very Poulsen, 2009), an increasing body of evidence suggests that different across Andean biomes (Pennington et al., 2010). While
1430 New Phytologist (2016) 210: 1430–1442 Ó 2016 The Authors www.newphytologist.com New Phytologist Ó 2016 New Phytologist Trust This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. New Phytologist Research 1431 high-elevation grassland clades are exceptionally young, fast, and interactions: fruit type and floral morphology. Abiotically dis- species-rich (e.g. Hughes & Eastwood, 2006), clades in lower- persed capsules characterize c. 40% of the species, while the elevation dry inter-Andean valleys are much older, slower, and remaining species produce fleshy, animal-dispersed berries, which species-poor (S€arkinen et al., 2012). Despite their species rich- evolved at least seven times from capsular-fruited ancestors ness, however, cloud forest biomes have been largely unexplored (Lagomarsino et al., 2014). Floral morphology is also highly vari- with respect to angiosperm diversification, although data suggest able. Lysipomia produces small, white flowers indicative of inver- that they are intermediate in terms of age and tempo (S€arkinen tebrate pollination (Faegri & van der Pijl, 1979), while both the et al., 2012). Chilean lobelias and the centropogonids produce robust flowers Identifying the causes of species-rich rapid radiations, such as that are usually vertebrate-pollinated. Flowers in the centropogo- those that characterize Andean high-elevation grasslands and nid clade are particularly diverse (Fig. 2) and principally adapted cloud forests, is a major goal in ecology and evolutionary biology. to pollination by hummingbirds and nectar bats (Colwell et al., While rapid diversification is undoubtedly the product of both 1974; Stein, 1992; Muchhala, 2006; Muchhala & Potts, 2007). abiotic and biotic factors acting on different spatial and temporal Owing to their broad Andean distribution and remarkable flo- scales (Antonelli & Sanmart ın, 2011; Ezard et al., 2011; Bouch ral and fruit diversity, the Neotropical bellflowers represent a enak-Khelladi et al., 2015), a tendency to explain a given diversity model group to examine the interaction of abiotic and biotic fac- pattern with a single process has persisted until recently. For tors that trigger mountain radiations. To accomplish this goal, montane Andean radiations, this was exemplified by invoking we develop and apply numerous statistical models to investigate molecular divergence time estimates contemporaneous with the diversification dynamics in the Neotropical bellflowers. More- final stages of Andean uplift, during which new habitats formed over, our analyses are among the first to incorporate geological and provided ecological opportunity for diversification (Hughes data into a model of evolutionary diversification (also see Valente & Eastwood, 2006; McGuire et al., 2014). However, other fac- et al., 2014 and Mulch, 2016). In this framework, we explicitly tors are probably at play simultaneously. For example, rapid plant investigate the influence of geology, climate, and ecological traits diversification in the region has also been attributed to specialized on species diversification using the first well-sampled, time- pollination relationships with hummingbirds (Schmidt-Lebuhn calibrated phylogeny of the group. et al., 2007; Abrahamczyk et al., 2014; Givnish et al., 2014), shifts between pollination syndromes (Kay et al., 2005), and the Materials and Methods acquisition of fleshy fruits in understory taxa (Smith, 2001; Givnish, 2010). Further, these factors are not likely to be inde- Taxon sampling and molecular dataset pendent of one another, e.g. mountain building affects the local climate (Armijo et al., 2015), which, in turn, alters biotic interac- One-third of Andean bellflower species were sampled, including tions (Blois et al., 2013). Not surprisingly, diversification studies all four species of Lobelia section Tupa, five of the c. 40 species of using molecular phylogenies have become more integrated in Lysipomia, and 191 species of the c. 550 species of the centro- their scope as methodology improves, and analyzing multiple fac- pogonid clade (Supporting Information Table S1). Our sampling tors underlying diversification patterns is becoming tractable includes representatives from all taxonomic subdivisions of (Ezard et al., 2011; Wagner et al., 2012; Givnish et al., 2014; Centropogon, Burmeistera, and Siphocampylus (Lagomarsino et al., Donoghue & Sanderson, 2015; Hughes et al., 2015). 2014). We assembled a concatenated molecular matrix that Here, we seek to untangle the ecological and historical factors includes seven plastid regions totaling 11 990 bp. Taxon sam- that have contributed to the generation of Andean megadiversity, pling, molecular methods for newly generated sequences, and especially in mesic mid-montane forests, using the Neotropical alignment protocols followed Lagomarsino et al. (2014). Out- bellflowers as a model system. The Neotropical bellflowers (Cam- group sampling includes broad representation across panulaceae: Lobelioideae) are an Andean-centered (Gentry, Lobelioideae. Nine representatives of Campanuloideae were used 1982) clade of c. 600 morphologically and ecologically diverse as outgroups to root the phylogeny and provide appropriate species found from lowland Amazonia to high-elevation grass- nodes for fossil calibration. lands above 5000 m, although the vast majority of species are found in cloud forests (Lagomarsino et al., 2014). The group was Phylogenetics and dating recently resolved into three well-supported subclades: the Chilean lobelias (Lobelia section Tupa; four species) of the temperate Phylogenies were inferred using maximum likelihood (ML) and southern Andes; the p aramo and puna endemic Lysipomia (50 Bayesian inference as implemented in RAxML 8.0 (Stamatakis, species); and the centropogonid clade of primarily cloud forest 2014) and BEAST 2.1.3 (Bouckaert et al., 2014). Four calibration endemics (Centropogon, Burmeistera, and Siphocampylus; c. 550 points were used to estimate divergence times: the fossil seed of species) (Lagomarsino et al., 2014). Lysipomia are diminutive †Campanula paleopyramidalis as a minimum age constraint of herbs, while the other subclades are robust and mostly woody. 16 Myr for the most recent common ancestor of C. pyramidalis The centropogonids are particularly diverse in their growth form: and C. carpatica (Cellinese et al., 2009; Crowl et al., 2014); a geo- most species are vines with woody bases, but herbs, shrubs, and logical maximum age constraint of 29.8 Myr, which corresponds trees are all represented (Fig. 1). Similar phenotypic diversity is to the age of the Kure atoll, the oldest emerged island of the also apparent in two traits known to facilitate plant–animal Hawaiian Ridge (Clague, 1996), for the crown group of the
Ó 2016 The Authors New Phytologist (2016) 210: 1430–1442 New Phytologist Ó 2016 New Phytologist Trust www.newphytologist.com New 1432 Research Phytologist
(a) (b) (c) (d) (e)
(f) (g) (h) (i)
Fig. 1 Growth form diversity in Neotropical bellflowers. (a) Siphocampylus tunarensis Zahlbr., a tree c.7–10 m. (b) Centropogon congestus Gleason, a clonal herbaceous species of wet soil, c. 1.5 m. (c) Centropogon pulcher Zahlbr., a hanging vine. (d) Burmeistera sp., a hemi-epiphytic herb. (e) Siphocampylus tunicatus Zahlbr., a tall shrub, with individual stems c. 5 m tall, arising from a single point. (f) Lysipomia muscoides Hook f., a minute rosette herb growing among moss in puna habitat. (g) Siphocampylus jelskii Zahlbr., a giant rosette shrub with elongated stems, each c.1–2 m tall, apparently clonal, growing in high-altitude grasslands. (h) Siphocamplus smilax Lammers, a xerophyte with a densely woody stem (probably water- storing) growing in sandstone. (i) Siphocampylus williamsii Rubsy, a plant producing a small number of narrow aerial stems arising from a xylopodium, or underground tuber-like stem. Photographs by: L. Lagomarsino (b–h), D. Santamar ıa-Aguilar (a), and A. Fuentes (i).
endemic Hawaiian clade, encompassing c. 125 spp. in six genera was assigned to the geological age constraint at 29.8 Myr (maxi- represented in our sampling by Brighamia, Cyanea, Clermontia mum hard bound), and normal priors were placed on the two and Delissea (Antonelli, 2009); and two secondary age constraints secondary age constraints (mean age of 56.0 Myr and SD of from Bell et al. (2010): 41–67 Myr ago for crown group Cam- 8.3 Myr for Campanulaceae; mean age of 43.0 Myr and SD of panulaceae, and 28–56 Myr ago for crown group Campanu- 8.0 Myr for Campanuloideae). The dated tree from treePL was loideae. The fossil record for Campanulaceae is very poor, and specified as the starting tree in each of eight separate BEAST runs, †C. paleopyramidalis, described from Miocene deposits in the which were each conducted for 10 million generations of Markov Nowy Sacz Basin in the West Carpathians of Poland (Łancucka- chain Monte Carlo (MCMC). Convergence was assessed using Srodoniowa, 1977), is the only fossil appropriate for calibration effective sample size (ESS) values of the runs in Tracer 1.6 of divergence time estimates in this family (Antonelli, 2009; (Rambaut et al., 2014), applying a cutoff value of 200. The maxi- Cellinese et al., 2009). It is assigned as a close relative of mum clade credibility tree, including credibility intervals (CIs) C. pyramidalis on the basis of their shared reticulate seed coats, a for ages and posterior probabilities (PPs) for node support, was feature uncommon in the genus (Łancucka-Srodoniowa, 1977; then assembled using TREEANNOTATOR (Bouckaert et al., 2014). Cellinese et al., 2009). Robustness of age estimates was assessed by removing one or a The optimal RAxML tree was dated using penalized likelihood series of calibration points. The following sets of calibration in treePL (Smith & O’Meara, 2012) with hard minimum and points were used for this purpose: the Campanulaceae secondary maximum age constraints. Confidence intervals were generated constraint only; the fossil and the Campanulaceae secondary age using 1000 RAxML bootstrap trees. We then simultaneously constraint; and the fossil and both secondary age constraints. inferred phylogenetic relationships and divergence times using Divergence time estimation can be sensitive to branch length both relaxed uncorrelated lognormal and exponential clock mod- variation, which is potentially influenced by growth form (Smith els in BEAST. A lognormal prior (mean of 1.5 and SD of 1.0) was & Donoghue, 2008). As they are generally herbaceous and have assigned to the fossil calibration age of 16 Myr, a uniform prior longer internal branches than their woody relatives, we suspected
New Phytologist (2016) 210: 1430–1442 Ó 2016 The Authors www.newphytologist.com New Phytologist Ó 2016 New Phytologist Trust New Phytologist Research 1433
Abiotic drivers of diversification 15 5000 (a) (c) 4500
10 4000
3500
3000 5
2500
2000 0 Elevation of tropical andes (m) Relative global temperature (°C)
1500
15 10 5 0 15 10 5 0
25 (b) 3.5 (d) 3.0 20
2.5
15 2.0
10 1.5
1.0 5 0.5
0 0.0
15 10 5 0 15 10 5 0
Speciation (blue) and Speciation (blue) (red) extinction rates Time (Myr) Time (Myr)
Extra-Andean 5 Low elevation (≤1900 m) 4 (e) Andean (f) High elevation (>1900 m) 4 3 3 2 2
1 1
0 0 Posterior probability density
0.0 0.5 1.0 1.5 0.0 0.5 1.0 1.5 Net diversification rate (events per Myr per lineage)
Biotic drivers of diversification 8 Invertebrate pollinated 5 (g) Capsules (h) Vertebrate pollinated Berries 4 6
3 4 2
2 1
0 Posterior probability pensity 0
0.0 0.5 1.0 1.5 0.0 0.5 1.0 1.5 Net diversification rate (events per Myr per lineage) 1
2 3
Middle Miocene Late Miocene Pliocene Pleistocene Andean Elevation Pollinator 15 10 5 0 Fruit type Fig. 2 Diversification of Neotropical bellflowers. A time-calibrated species-level phylogeny shows the c. 5 Myr age of the largely Andean centropogonid clade (c. 550 species, node 1), whose origin is associated with a significant increase in diversification rate (yellow star) as detected by Bayesian Analysis of Macroevolutionary Mixture (BAMM). Lysipomia (c. 50 species) and Chilean Lobelia (four species) are indicated (nodes 2 and 3, respectively). Representative floral diversity, shown on the right, illustrates the striking phenotypic diversity in the clade (see Fig. 1 for comparable growth form diversity); scale bars, 0.5 cm. Diversification models examining the abiotic correlates of this rapid diversification include average paleoelevation of the tropical Andes through time (a) and global temperatures through time (c). Gray dots in (a) and (c) represent individual data points utilized to create the curves. Results from these models show inferred speciation (blue) and extinction (red) rates through time under models depending on paleoelevation (b) and paleotemperature (d). Additional diversification analyses using binary state-speciation and extinction (BiSSE) demonstrate the effect of two abiotic and two biotic traits on net diversification rate: Andean occurrence (e; extra-Andean (red) vs Andean (blue)), elevation (f; low elevation, ≤ 1900 m (orange) vs high elevation, > 1900 m (purple)), fruit type (g; dry capsules (light blue) vs fleshy berries (pink)), and pollinator type (h; invertebrate (yellow) vs vertebrate (green)). Trait scorings are color-coded to the right of phylogeny. Outgroups were removed, and taxon names omitted because of space constraints. Photographs by L. Lagomarsino. Geological timescale shown at bottom from Walker et al. (2012).
Ó 2016 The Authors New Phytologist (2016) 210: 1430–1442 New Phytologist Ó 2016 New Phytologist Trust www.newphytologist.com New 1434 Research Phytologist
this may be the case for Lysipomia. We thus removed Lysipomia parameters for each model. The model supported by the lowest species entirely, and re-estimated molecular divergence times to AICc was considered best if it was at least two ΔAIC units better determine if these differences had an impact on results. We find than the model with the second lowest score. We also used likeli- no evidence that the results are affected by dating method, hood ratio tests (significant at P < 0.05) to estimate support for calibration strategy, or branch length heterogeneity: the 95% the best-fitting model when compared with rival models. CIs of the re-estimated ages overlap with the 95% CIs from the The diversification models we apply assume that extinction BEAST analysis that we use for diversification analyses (see also rates can be estimated from molecular phylogenetic branching Table S2). patterns, as originally proposed by Nee et al. (1994). It is impor- tant to note this is a controversial practice (Rabosky, 2010, 2015), especially without incorporating paleontological data Trait categorization (Quental & Marshall, 2010). Despite this, recent studies have Species traits were newly obtained from a variety of sources, suggested that unbiased estimates of extinction rates can be including: online databases (GBIF, http://www.gbif.org/ and obtained solely from molecular phylogenies (Morlon et al., 2011; Tropicos, http://www.tropicos.org/), herbarium specimens, taxo- Beaulieu & O’Meara, 2015). In this spirit, we explore the indi- nomic literature (Wimmer, 1943, 1953), and field collections. vidual components of diversification rate in our discussion of Morphological characters and area of occurrence were coded results, particularly with respect to paleoenvironmental change. from type specimens available on the JSTOR Global Plants database (http://plants.jstor.org/) for nearly all 594 species, many Time-dependent diversification of which were not represented in our phylogenetic sampling but were used to account for sampling biases. We scored four binary We used BAMM to estimate speciation and extinction rates traits related to the biology and geographic occurrence of all through time and to identify shifts in diversification rate. We Andean bellflower species for our diversification analyses: fruit accounted for incomplete taxon sampling by applying clade- type (berry or capsule), pollinator type (vertebrate or inverte- specific sampling fractions in each of the three principal Andean brate), elevation (median species occurrence ≤ 1900 m vs bellflower subclades: the Chilean lobelias (1.0), Lysipomia (0.12), > 1900 m), and presence/absence in the Andes. Pollinator type and the centropogonids (0.35). We were unable to provide finer- was coded according to classical morphological definitions of pol- scale sampling fractions within the centropogonids because of lination syndromes (Faegri & van der Pijl, 1979; Muchhala, nonmonophyly of genera and the difficulty of placing species into 2006). Because they are not sensitive to outliers, we used median their constituent clades, particularly within Siphocampylus (Lago- elevation across all specimen locality data to determine each marsino et al., 2014). While our taxon sampling is fairly low, species value for elevation. We further used the global median of recent studies exploring the effect of incomplete taxon sampling these values across species as our cutoff for high vs low elevation on the identification of rate shifts suggest that, while results may because this resulted in an equal distribution of the character be incomplete, they should be accurate (Spriggs et al., 2015). We states (high vs low) across the phylogeny, which is ideal for binary ran BAMM with four reversible jump MCMC chains, each for state-speciation and extinction (BiSSE) analyses (Maddison & five million generations. ESS values (> 200) were used to assess FitzJohn, 2015). All trait data are provided in Table S3 and in convergence. The posterior distribution was used to estimate the Fig. 2 and are deposited in Dryad (doi: 10.5061/dryad.7h4j). configuration of the diversification rate shifts, and alternative diversification models were compared using Bayes factors. Results were analyzed and plotted using the R package BAMMTOOLS Diversification analyses 2.0.2 (Rabosky et al., 2014). We applied a series of birth–death models to quantify the effects of abiotic and biotic correlates of speciation and extinction in Paleoenvironment-dependent diversification Neotropical bellflowers. We assessed the robustness of our results in a variety of ways. First, to accommodate phylogenetic and dat- We quantified the effect of the past environment (i.e. climate ing uncertainties (except in the case of the diversity-dependent change and Andean surface uplift) on speciation and extinction analyses, which are computationally intensive), we conducted our rates in Neotropical bellflowers via an approach that builds on analyses across 500 randomly sampled trees from the BEAST time-dependent diversification models (Nee et al., 1994; Morlon posterior distribution (outgroups removed). Second, we imple- et al., 2011). The method we utilized allows speciation and mented a sampling fraction to account for nonsampled species extinction rates to correlate not only with time, but also with a diversity. We also applied a sampling fraction, including trait quantitative external variable that is time-dependent (see Con- data for species not sampled in our phylogeny, to account for any damine et al., 2013 for details). These paleoenvironment- sampling bias in trait space for BiSSE analyses (Maddison et al., dependent birth–death models are implemented in the R package 2007). Third, for each of our diversification analyses except for RPANDA (Morlon et al., 2016). We incorporated two quantitative those using Bayesian Analysis of Macroevolutionary Mixture external paleoenvironmental variables: the global Cenozoic deep- (BAMM 2.2.2, Rabosky, 2014), we selected the best-fitting sea oxygen isotope record as a proxy for global temperature model by computing the corrected Akaike information criterion (Zachos et al., 2008), and a generalized model of the paleoeleva- (AICc) based on the log-likelihood and the number of free tion history of the tropical Andes. The latter was compiled from
New Phytologist (2016) 210: 1430–1442 Ó 2016 The Authors www.newphytologist.com New Phytologist Ó 2016 New Phytologist Trust New Phytologist Research 1435 several references (Garzione et al., 2006, 2008, 2014; Ehlers & particular traits on species diversification, particularly when these Poulsen, 2009; Leier et al., 2013) and is available on the Dryad caveats have been mitigated (Ng & Smith, 2014), as we have repository (doi: 10.5061/dryad.7h4j). attempted to do here. Where possible, we also compare the trait- The R package PSPLINE was used to interpolate a smooth line dependent results with the inferences made with nontrait- for each environmental variable. This smooth line was sampled dependent models (i.e. BAMM and RPANDA) to show that that during each birth–death modeling process to give the value of the they are consistent. paleoenvironmental variable at each time point. Speciation and/ or extinction rates were then estimated as a function of these Diversity-dependent diversification values along the dated phylogenies according to the parameters of the following models. For both temperature and Andean We applied models that allow speciation and extinction to vary elevation, we first applied two standard models, i.e. constant according to the number of lineages in the phylogeny. Using the speciation without extinction (Yule model) and constant-rate R package DDD 2.7 (Etienne et al., 2012), we built two models: birth–death, as null references for model comparison. We then speciation declines with diversity, assuming no extinction, and applied four models in which the paleoenvironment dependency speciation declines with diversity, allowing for extinction. The ini- was exponential (Table S4): speciation rate varies with the envi- tial carrying capacity was set to the current species diversity, and ronmental variable and no extinction; speciation rate varies with the final carrying capacity was estimated according to the model the environmental variable and extinction rate is constant; specia- and corresponding parameters. These models were run on the tion rate is constant and extinction rate varies with the environ- maximum clade credibility (MCC) tree from our BEAST analysis. mental variable; and both speciation and extinction rates vary with the environmental variable. We repeated these models with Results linear dependence on the environmental variable (Table S4). Molecular phylogeny and divergence time estimation Trait-dependent diversification The final alignment for phylogenetic analysis comprised seven We modeled the impact of traits on the diversification of loci (c. 12 000 bp) sequenced for 275 species (doi: 10.5061/ Neotropical bellflowers by concurrently estimating their impact dryad.7h4j) (Table S1). The concatenated Bayesian phylogenetic on speciation, extinction, and transition rates using BiSSE (Mad- analysis using the best-fitting partitioning scheme yielded a highly dison et al., 2007). For each of the four traits (Andean occur- resolved and strongly supported tree compatible with the ML tree rence, elevation, pollination syndrome, and fruit type), we and previously published results (Lagomarsino et al., 2014). evaluated eight BiSSE diversification models of increasing com- Sequence divergence, particularly in the centropogonid clade, was plexity in which speciation, extinction, and transition rates were low, as reflected by the shallow branching pattern toward the tips. allowed to either vary or remain equal between traits (Table S5). Despite short internal branches, indicative of a recent radiation, Analyses were performed using the R package DIVERSITREE 0.7-6 the phylogeny is generally well supported: 66% of nodes are (FitzJohn, 2012). Once the best-fitting model was selected, CIs recovered with moderate support (PP > 0.90) and 28% are well for each parameter were estimated for the tree. We used an expo- supported (PP > 0.95). Our results support the monophyly of the nential prior following FitzJohn (2012), and began the MCMC subfamilies Campanuloideae and Lobelioideae and the Neotropi- with the ML estimates. We ran MCMC for 20 000 generations cal bellflowers (the centropogonids, Lysipomia, and Chilean and applied a burn-in of 2000 steps. We then computed the net lobelias) with maximum nodal support (PP = 1). diversification rates for each trait. Finally, to determine if these Results of the dating analyses are presented in Fig. 2 (see Figs traits were correlated, pairwise trait comparisons were performed S1 and S2 for the 95% CI for each node). Similar estimates were across a random sampling of 500 trees from the posterior distri- obtained using different dating methods and with alternative cali- bution in both a ML and Bayesian framework in BAYESTRAITS 2.0 bration scenarios and settings (Table S2). This suggests that (Pagel & Meade, 2006). The significance of binary trait depen- potential biases such as long internal branch lengths, possibly dence was assessed against a rival model where these traits evolved attributed to growth form variation, do not have a large effect on independently using likelihood ratio tests (ML) and Bayes factors our age estimates. Based on the molecular divergence time esti- of the harmonic mean of likelihood values. mates from BEAST (Bouckaert et al., 2014) and treePL (Smith & Binary state-speciation and extinction models have received O’Meara, 2012), we place the origin of the extant Neotropical recent criticism, including high type I error rates (Maddison & bellflowers diversity in the Miocene, c. 17.1 Myr ago (95% CI: FitzJohn, 2015; Rabosky & Goldberg, 2015), especially for trees 14.32–20.32 Myr ago). We further found that the crown groups with fewer than 300 terminals and for traits present in < 10% of of their three main subclades – Chilean lobelias, Lysipomia, and taxa (Davis et al., 2013). The distribution of characters states is the centropogonids – originated 1.45 Myr ago (95% CI: 0.42– also important; for example, rapid diversification rates in a region 2.78 Myr ago), 11.87 Myr ago (95% CI: 9.00–15.24 Myr ago), of a phylogeny can be erroneously attributed to a particular char- and 5.02 Myr ago (95% CI 3.95–6.13 Myr ago), respectively acter state when the trait is characterized by high transition rates (Figs 2, S1). All branching events in the most species-rich clade, (Rabosky & Goldberg, 2015). However, despite these known the centropogonids, occur in the Plio-Pleistocene (5.3–0.01 Myr issues, SSE models remain a viable framework to test the effect of ago).
Ó 2016 The Authors New Phytologist (2016) 210: 1430–1442 New Phytologist Ó 2016 New Phytologist Trust www.newphytologist.com New 1436 Research Phytologist
The age inferred for Lysipomia is surprisingly old considering distribution (high elevation (> 1900 m) vs low elevation that the genus is currently restricted to p aramo and puna vege- (≤ 1900 m)). The best-fitting model for these BiSSE analyses tations, the youngest of Andean biomes (van der Hammen & inferred significantly higher speciation rates for Andean species Cleef, 1986). This older age is potentially an artifact caused by and for species occurring at high elevation (Figs 2e,f, S5; the substantial DNA substitution rate heterogeneity observed Table S5), respectively. The similar results of these BiSSE between Lysipomia and the rest of the phylogeny, as illustrated analyses are potentially due to the correlated nature of eleva- in the phylogram in Fig. S3. This elevated substitution rate may tional distribution and Andean occurrence that we identified be explained by Lysipomia’s herbaceous habit or high-altitude in BAYESTRAITS (Pagel & Meade, 2006) (Table S6); this corre- occurrence (Lagomarsino et al., 2014). It is likely that this sub- lation is not surprising given that the Andes represent the stantial rate heterogeneity is not adequately accommodated by majority of the topographic relief across the Neotropical the rate smoothing in BEAST, as documented in other groups bellflowers’ range. Together, these results are consistent with that exhibit between-clade rate variation (Beaulieu et al., 2015). an integral role of Andean surface uplift as a trigger of rapid The age estimates for Lysipomia are thus potentially overesti- diversification for Neotropical bellflowers. mated as a result, which may impact diversification analyses. Using the same paleoenvironmental birth–death model, but However, we have shown that this rate heterogeneity does not instead incorporating relative global temperature proxies that affect the estimated age of the other two major subclades of span the last c. 20 Myr of the Cenozoic (Zachos et al., 2008), Neotropical bellflowers, and as such the impact should be we demonstrate a significant correlation between climate change minimal (Table S2). and Andean bellflower diversification (Fig. 2d). In contrast to Andean orogeny, which influenced primarily speciation, the best-fitting paleoclimatic model suggests that climate change had Diversification analyses a significant impact only on extinction. Beginning c. 15 Myr ago A time-dependent diversification analysis in BAMM strongly (mid-Miocene), extinction rates decreased dramatically towards rejects constant diversification. Instead, it indicates that the the present as a function of global temperature (Fig. 2d; Neotropical bellflowers underwent a significant shift in net Table S4). diversification rate coinciding with the recent origin of the cen- Additional BiSSE analyses demonstrate that various clade- tropogonid subclade (c. 5 Myr ago), which includes 550 of the specific biotic traits influence diversification rates in Neotro- 600 species (Figs 2, S4). This analysis further provides evidence pical bellflowers. BAYESTRAITS analyses showed uncorrelated for a 3.7-fold increase in diversification rate (1.83 vs 0.5 evolution of these traits with respect to one another, and with – events Myr 1 per lineage); net diversification rate is similarly both the abiotic factors discussed earlier (Table S6). The best high using an alternative metric (Magallon & Sanderson, 2001) models, as supported by both AICc and likelihood ratio tests – (0.82–1.15 events Myr 1 per lineage under high (e = 0.90) and (Table S5), infer higher net diversification rates in lineages with low (e = 0) extinction fractions, respectively). As illustrated in two character states: fleshy fruits (vs dry fruits; Figs 2g, S5), Fig. S4(b), the diversification rate for the centropogonid clade is and vertebrate pollination syndromes (vs invertebrate pollina- much higher than its background rate. tion syndromes; Figs 2h, S5). These traits acted differently on We then adapted a recently developed paleoenvironment- the two components of diversification: fleshy fruits confer lower dependent diversification model that explicitly accommodates extinction rates than capsules, while species pollinated by verte- the effect of changing environment through time (Condamine brates have higher speciation rates than invertebrate-pollinated et al., 2013) to examine the effect of Andean paleoelevation species. (Fig. 2a) and past temperature on diversification rate (Fig. 2d). Finally, a diversity-dependent analysis inferred that the This novel integration of Andean geological data into an anal- Neotropical bellflowers have not reached diversity equilibrium: ysis of evolutionary diversification reveals that speciation rates the estimated carrying capacity (i.e. number of species potentially in the Neotropical bellflowers are positively correlated with sustained in the clade) far exceeds the current extant diversity of the elevation history of the Andes, while extinction rates the clade (Table S7). appear to be unlinked to mountain uplift (Fig. 2b; Table S4). Here, we reconstruct a continuous increase in speciation rate Discussion beginning at c. 12 Myr ago (mid-Miocene) and peaking at c. 5 Myr ago (early Pliocene), consistent with our result from The Andes are famous for their high species richness (Myers BAMM. This coincides with the attainment of maximum et al., 2000) and the rapid diversification rates that characterize paleoelevation of the Andes (> 4000 m), at which point we many of their emblematic clades (Hughes & Eastwood, 2006; – infer an unprecedented ~15 speciation events Myr 1 per lin- Madrin~ an et al., 2013; Givnish et al., 2014). Although numer- eage for Neotropical bellflowers. Because of the magnitude of ous studies have documented this pattern, there have been few this result, we ran additional analyses implementing a trait- attempts to ascertain the processes that underlie this megadi- dependent diversification model using BiSSE (Maddison et al., versity, especially in the cloud forests where Neotropical 2007). We investigated two traits that serve as a proxy for the bellflower diversity is concentrated. Our study tackles this issue effect of Andean orogeny on diversification in this framework: with an interdisciplinary approach that paves the way for Andean presence (Andean vs extra-Andean) and elevational future inquiry.
New Phytologist (2016) 210: 1430–1442 Ó 2016 The Authors www.newphytologist.com New Phytologist Ó 2016 New Phytologist Trust New Phytologist Research 1437
hypothesize that Neotropical bellflowers evolved their tolerance The impact of Andean orogeny and climate change on to cool habitats in southern, perhaps temperate, regions of South species diversification America, similar to where Chilean lobelias now occur, before Within the morphologically and ecologically diverse Neotropical major tropical Andean surface uplift. This climatic preadaptation bellflower clade, a shift to increased diversification rates on the may explain their northward expansion and decreased extinction branch subtending the centropogonid subclade resulted in a cloud rates as this clade exploited cooler niches associated with the ris- forest-centered radiation with c. 550 extant species. The diversifi- ing Andes. These results significantly improve our understanding cation rate for the centropogonids, estimated to be between 0.65 of the abiotic determinants that have triggered Andean diversifi- – and 1.42 events Myr 1 per lineage, makes them one of fastest cation, and collectively indicate that mountain building and cli- species radiations reported to date (Table 1), with average overall mate are key contributors to increases in diversification rate in rates of speciation exceeding iconic groups ranging from Espeletia Neotropical bellflowers, although they have acted on different (Madrin~ an et al., 2013) and hummingbirds (McGuire et al., components of net diversification. 2014) in the Andes to the silverswords (Baldwin & Sanderson, 1998) and Drosophilidae of Hawaii (Lapoint et al., 2014). Biotic traits that facilitate mutualisms are associated with In addition, our analyses provide key insights into the abiotic elevated rates of net diversification and biotic drivers of rapid diversification in the Andes. Using diversification models that incorporate geological data, we In addition to orogeny and climate change, our trait-dependent demonstrate that mountain building is correlated with increased analyses point to a role of species ecology in influencing species diversification rates. Specifically, we demonstrate that the final diversification in Neotropical bellflowers (Table S5). Specifically, stages of Andean uplift are strongly associated with the origin we identify that mutualistic interactions with seed dispersers and and rise of prolific speciation rates in the Neotropical bellflowers pollinators promote diversification. Here, we demonstrate that (Fig. 2b; Table S4). These rates peaked at a remarkable c.15 bird-dispersed berries and vertebrate-pollinated flowers are associ- – events Myr 1 per lineage, concurrent with the attainment of the ated with significantly higher rates of diversification than their maximum past elevations of the Andes. Although not sustained, counterparts (Figs 2g,h, S4; Table S5). Species with berries have a this diversification rate exceeds the highest reported in plants, net diversification rate that is c. 3.5 times higher than capsular – including in the Mediterranean carnations (2.2–7.6 events Myr 1 species, whose seeds are abiotically dispersed (Figs 2g, S4). As in per lineage, Valente et al., 2010) and Andean lupines (2.49–5.21 other groups, fleshy fruits in Neotropical bellflowers are generally – events Myr 1 per lineage, Hughes & Eastwood, 2006; Drum- associated with densely forested understory habitats where abiotic mond et al., 2012). These results are corroborated by the shift to dispersal is not as effective as in open habitats, where capsular taxa faster speciation rates that was detected by BAMM along the dominate (Lagomarsino et al., 2014). Owing to the limited move- branch that subtends the centropogonid clade, during the period ment of understory birds, dispersal of fleshy fruits in these habi- of final Andean uplift. Furthermore, BiSSE analyses bolster our tats is typically across very short distances (Theim et al., 2014). findings that the Andes in general are a major driver of species When coupled with rare long-distance dispersal events, these diversification in this clade: Andean taxa have significantly higher short dispersal distances can lead to rampant allopatric speciation net diversification rates than extra-Andean taxa (Figs 2e, S5; (Givnish et al., 2005, 2009; Givnish, 2010). Further, the repeated Table S5). Moreover, species diversification is faster at higher ele- evolution of berries in the centropogonid clade may be indicative vations (> 1900 m) (Figs 2f, S5: Table S5). The likely driver of of multiple parallel radiations across the Andes, each with rapid this impressive diversification was not paleoelevation per se (as speciation rates, consistent with a model of cordilleran diversifica- would be expected under a predominantly vicariant model of tion, as recently proposed in bromeliads (Givnish et al., 2014). speciation), but rather ecological opportunity triggered by the Pollination by vertebrates appears to have played an even big- availability of wet montane forests. These habitats emerged and ger role in species diversification than seed dispersal. Here, we see greatly expanded across the eastern flanks of the Amazon in the an approximately sixfold increase in diversification rate relative to late Miocene and early Pliocene during periods of active moun- invertebrate pollination (Figs 2h, S5; Table S5). Vertebrate polli- tain building (van der Hammen & Cleef, 1986). nators, namely hummingbirds and bats, are hypothesized to be We further present evidence that past climate change has inde- more effective pollinators than invertebrates, in part because they pendently influenced diversification rates in the Neotropical carry higher pollen loads over longer distances (Fleming et al., bellflowers. Our paleotemperature-dependency analysis shows 2009). The increased speciation rates among vertebrate- that, while speciation rates remained high but constant, extinc- pollinated taxa (Table S5; Fig. S5) are a likely result of floral iso- tion rates are positively correlated with past temperature. Conse- lation, which allows prezygotic reproductive isolation to be quently, inferred extinction rates were lowest during the coolest achieved via an interplay of floral morphology and pollinator past intervals, and have generally been declining since the origin behavior (Fulton & Hodges, 1999; Muchhala, 2003; Schiestl & of the clade, a result that is particularly striking in today’s world Schluter,€ 2009). Supporting this, many Andean bellflower species of anthropogenic warming. Although our proxy of climate have sympatric distributions that appear to have been facilitated change is global, it is almost certain that cooling since the mid- by either shifts between different classes of vertebrate pollinators Miocene has affected the Andes, especially at mid-to-high eleva- (Muchhala, 2006) or character partitioning of floral traits in tions, where Neotropical bellflowers are most diverse. We species that share pollinators (Muchhala & Potts, 2007).
Ó 2016 The Authors New Phytologist (2016) 210: 1430–1442 New Phytologist Ó 2016 New Phytologist Trust www.newphytologist.com 1438 Research e Phytologist New www.newphytologist.com (2016) 210: 1430–1442 Table 1 Net diversification rates in the fastest species radiations in the Tree of Life, including the centropogonids
Species Clade Area Organism richness Age (95% CI) r (e = 0) r (e = 0.5) r (e = 0.9)
Dianthus (carnations) (Valente et al., 2010) Eurasia Plant 200 1.5 (0.9–2.1) 3.07 (2.12–5.12) 2.88 (2.06–4.80) 1.99 (1.42–3.32) African cichlids (Salzburger et al., 2005) East African Fishes 1800 2.4 (1.22–4.02) 2.83 (1.69–5.76) 2.71 (1.62–5.34) 2.14 (1.28–4.22) Lakes Lupinus (Andean clade) (Hughes & Eastwood, 2006) Andes Plant 81 1.47 (1.18–1.76) 2.52 (2.10–3.14) 2.33 (1.95–2.90) 1.46 (1.22–1.82) Neo-astragalus (clade F) (Scherson et al., 2008) Andes Plant 90 1.89 (1.71–2.07) 2.04 (1.84–2.23) 1.87 (1.70–2.06) 1.19 (1.08–1.31) Zosterops (Moyle et al., 2009) Cosmopolitan Birds 80 1.84 (1.40–1.89) 2.0 (1.95–2.63) 1.85 (1.80–2.44) 1.16 (1.13–1.52) Neo-astragalus (clade G) (Scherson et al., 2008) Andes Plant 15 0.98 (0.79–1.17) 2.06 (1.72–2.55) 1.82 (1.52–2.26) 0.84 (0.70–1.04) Gentianella (von Hagen & Kadereit, 2001) Andes Plant 170 1.6–3.0 1.48–2.78 1.39–2.6 0.94–1.77 Cistus (Guzman et al., 2009) Mediterranean Plant 10 1.04 (0.66–1.29) 1.54 (1.25–2.44) 1.35 (1.09–2.13) 0.57 (0.46–0.89) Paussus (ant-nest beetles) (Moore & Robertson, 2014) Madagascar Insects 86 2.6 (1.7–3.6) 1.45 (1.04–2.21) 1.34 (0.97–2.05) 0.85 (0.61–1.29) Centropogonids (this study) Andes Plant 556 5.02 (3.95–6.13) 1.12 (0.91–1.42) 1.06 (0.87–1.35) 0.79 (0.65–1.00) Espeletiinae (Madrin~an et al., 2013) Andes Plant 120 4.04 (2.42–5.92) 1.01 (0.69–1.69) 0.944 (0.64–1.58) 0.62 (0.42–1.04) AMC clade of Hawaiian Drosophila (Lapoint et al., 2014) Hawaii Insect 91 4.40 (3.45–11.82) 0.94 (0.35–1.2) 0.88 (0.33–1.12) 0.58 (0.22–0.74) Inga (Richardson et al., 2001) Neotropics Plant 300 5.9 (2.0–13.4) 0.85 (0.37–2.5) 0.8 (0.35–2.36) 0.57 (0.25–1.69) Ruschioideae (Klak et al., 2003) South Africa Plant 1563 3.8–8.7 0.7–1.75 0.73–1.68 0.58–1.31 Bee hummingbirds (McGuire et al., 2014) Andes Birds 36 5 0.58 0.53 0.29 Silversword alliance (Baldwin & Sanderson, 1998) Hawaii Plants 28 5.2 (4.4–6.0) 0.51 (0.44–60) 0.46 (0.40–0.54) 0.24 (0.21–0.29) Costus subgenus Costus (Kay et al., 2005) Neotropics Plant 51 1.5–7.1 0.46–2.16 0.42–1.98 0.24–1.16 Lupinus ‘super radiation’ (Drummond et al., 2012) Western Plant 196 5.0–13.2 0.35–0.92 0.33–0.86 0.22–0.59 Americas Hawaiian lobeliads (Givnish et al., 2009) Hawaii Plants 126 13.6 (10.49–16.71) 0.30 (0.25–0.39) 0.28 (0.23–0.37) 0.19 (0.15–0.24) Hawaiian Drosophilidae (Lapoint et al., 2014) Hawaii Insect 1000 25.15 (23.90–27.46) 0.25 (0.23–0.26) 0.24 (0.22–0.25) 0.18 (0.17–0.19) Plethodon, glutinosus group (Kozak et al., 2006) North America Amphibian 30 8–11 0.24–0.34 0.22–0.31 0.12–0.16 e Phytologist New Hummingbirds (all) (McGuire et al., 2014) Neotropics Bird 338 22.4 (20.3–24.7) 0.23 (0.21–0.25) 0.22 (0.20–0.24) 0.16 (0.14–0.17) Ctenotus + Lerista (Rabosky et al., 2007) Australia Reptile 174 c. 20 0.22 0.21 0.14
Rates of diversification as calculated using the statistic of Magallon & Sanderson (2001) under low, medium, and high estimated extinction (e) for the centropogonid clade (bold text) rank ordered – with other published rapid radiations. All rates in events Myr 1 per lineage. Estimates derived from reported ages and species richness from the relevant literature are cited. When possible, 95% confi-
Ó dence intervals (CIs) for ages are reported, except in cases where sufficient data were unavailable. 06NwPyooitTrust Phytologist New 2016 Ó 06TeAuthors The 2016 Phytologist New New Phytologist Research 1439
Our analyses of biotic traits indicate that ecological factors, clades, it is an additive interaction of ecological and environ- particularly mutualisms with other species, constitute additional mental factors underlies this region’s megadiversity. triggers of rapid diversification rates in Neotropical bellflowers, More generally, rapid plant diversification in the world’s mon- as has been demonstrated in extra-Andean plant groups (Weber tane regions is frequently a product of island-like ecological & Agrawal, 2014). This effect of species ecology is especially pro- opportunity following mountain uplift, further stimulated by nounced within the centropogonids, in which more than half of evolutionary innovation (Hughes & Eastwood, 2006; Hughes & the species are vertebrate-pollinated and/or produce berries, and Atchison, 2015). Our analyses demonstrate that this model in which diversification rates are particularly elevated (Fig. S4). applies to Neotropical bellflowers, and is likely relevant to other Our results further demonstrate that diversity-dependent pro- Andean cloud forest clades. However, why are the Andes more cesses did not govern diversification, suggesting that the clade is diverse than other montane systems? It is likely that the recency in its early phases of diversification and is not presently bounded of Andean mountain building partially explains this phe- by ecological limits. nomenon: many clades may still be in the early phases of explo- sive radiation and thus are not yet greatly impacted by extinction. In addition, the Andes are characterized by deeply dissected A synergy of abiotic and biotic factors drives diversification topography, including some of the steepest environmental gradi- in Andean cloud forests ents known (Hughes & Atchison, 2015). These characteristics, Although the interplay between abiotic and biotic drivers has which almost certainly enhance the island-like nature of this sys- long been recognized as fundamental for regulating diversity, tem and further intensify ecological opportunity, are shared with progress toward understanding their interaction has been slow. the Hengduan Mountains (Favre et al., 2014), another of the Our study is among the first to explore this interaction in a world’s biodiversity hotspots (Myers et al., 2000; Kreft & Jetz, clade distributed predominantly in Andean cloud forests. Using 2007). We are likely to gain insight into why the Andes are so the Neotropical bellflowers as a model system, we first provide diverse by comparing their clades’ diversification histories with important evidence that the age and tempo of diversification of those in other species-rich montane regions. Our framework, cloud forest clades are intermediate between those in the older including the integration of climatic, paleoaltimetry, and trait inter-Andean valleys and more recent high-elevation grasslands data, can be used towards this end. (Pennington et al., 2010; S€arkinen et al., 2012). We then demonstrate that Andean surface uplift and climate are both Acknowledgements significantly correlated with the rapid radiation of Neotropical bellflowers. In the context of Andean biodiversity, dramatic We thank members of the Davis and Antonelli research groups geological and environmental transformations, often resulting for helpful feedback, and Suzette Flantua, Elisabeth Forrestel, in habitat barriers and steep environmental gradients (e.g. high Carina Hoorn, Daniel Santamar ıa-Aguilar, and Zhenxiang Xi for ridges and low valleys), characterized the matrix in which plant insightful conversations. The directors and curators at the follow- lineages and their mutualists diversified. Our results indicate ing herbaria provided important access to their collections: that key differences in biotic traits, including fruit type and BOLV, CR, GB, GH, INB, LPB, MO, MOL, NY, PMA, SCZ, pollination syndrome, confer enhanced diversification capacity SMF and USZ (acronyms follow the Index Herbariorum, Theirs among closely related species co-occurring in this rapidly et al., 2013 (continuously updated)). We are also very thankful changing environment. We hypothesize that preadaptation to for the detailed and insightful comments from two anonymous cooler climates as the tropical Andes rose allowed this clade to reviewers and the editors. Funding was provided by a National achieve its broad Andean distribution, centered in wet, mid- Science Foundation Doctoral Dissertation Improvement Grant elevation cloud forests. This, in turn, precipitated mutualistic DEB-1210401 to L.P.L. and C.C.D.; by the Swedish Research interactions with groups showing similarly high Andean diver- Council (B05 69601), the European Research Council under the sity (e.g. hummingbirds), further enhancing diversification. European Union’s Seventh Framework Programme (FP/2007- Within the context of fruit type, limited dispersibility coupled 2013, ERC Grant Agreement no. 331024), and a Wallenberg with rare long-distance dispersal events in berry-producing lin- Academy Fellowship to A.A.; by the Carl Tryggers Stiftelse (CTS eages resulted in increased diversification rates, probably via 12:14) and a Marie Curie Actions (BIOMME project, IOF- allopatric speciation, as shown in the closely related Hawaiian 627684) to F.L.C.; and by the Department of Organismic and lobeliads (Givnish et al., 2009). At the same time, an associa- Evolutionary Biology, the Botanical Society of America, the tion with vertebrate pollinators resulted in increased rates of American Society of Plant Taxonomists, the Arnold Arboretum’s speciation via floral isolation. As also demonstrated in bromeli- Deland Award, the Explorer’s Club, and the Rockefeller Center ads (Givnish et al., 2014), the repeated evolution of fruit types for Latin American Studies to L.P.L. and pollination syndromes in the complex landscape of the Andes led to multiple parallel radiations in Neotropical Author contributions bellflowers, which together resulted in a very large, very fast radiation: the centropogonid clade. We argue that while the L.P.L., F.L.C., A.A. and C.C.D. designed the study; A.M. com- Andes, and the habitats and habitat heterogeneity created by piled the paleoaltimetry data; L.P.L. assembled all other data; their orogeny, acted as a montane species pump across many L.P.L. and F.L.C. analyzed data; and L.P.L. and C.C.D. wrote
Ó 2016 The Authors New Phytologist (2016) 210: 1430–1442 New Phytologist Ó 2016 New Phytologist Trust www.newphytologist.com New 1440 Research Phytologist
the paper, with significant contributions from F.L.C. and A.A. Elias M, Joron M, Willmott K, Silva-Brand~ao KL, Kaiser V, Arias CF, Gomez ~ et al. and revisions from A.M. Pinerez LM, Uribe S, Brower VZ, Freitas VL 2009. Out of the Andes: patterns of diversification in clearwing butterflies. Molecular Ecology 18: 1716– 1729. References Etienne RS, Haegeman B, Stadler T, Aze T, Pearson PN, Purvis A, Phillimore AB. 2012. Diversity-dependence brings molecular phylogenies closer to Abrahamczyk S, Souto-Vilar os D, Renner SS. 2014. Escape from extreme agreement with the fossil record. Proceedings of the Royal Society of London. specialization: passionflowers, bats and the sword-billed hummingbird. Series B: Biological Sciences 279: 1300–1309. Proceedings of the Royal Society of London. Series B: Biological Sciences 281: Ezard THG, Aze T, Pearson PN, Purvis A. 2011. Interplay between changing 20140888. climate and species’ ecology drives macroevolutionary dynamics. Science (New Antonelli A. 2009. Have giant lobelias evolved several times independently? Life York, N.Y.) 332: 349–351. form shifts and historical biogeography of the cosmopolitan and highly diverse Faegri K, van der Pijl L. 1979. The principles of pollination ecology. Oxford, UK: subfamily Lobelioideae (Campanulaceae). BMC Biology 7: 82. Pergamon Press. Antonelli A. 2015. Biodiversity: multiple origins of mountain life. Nature 524: Favre A, P€ackert M, Pauls SU, J€ahnig SC, Uhl D, Michalak I, Muellner-Riehl 300–301. AN. 2014. The role of the uplift of the Qinghai-Tibetan Plateau for the Antonelli A, Sanmart ın I. 2011. Why are there so many plant species in the evolution of Tibetan biotas. Biological Reviews 90: 236–253. Neotropics? Taxon 60: 403–414. FitzJohn RG. 2012. Diversitree: comparative phylogenetic analyses of Armijo R, Lacassin R, Coudurier-Curveur A, Carrizo D. 2015. Coupled tectonic diversification in R. Methods in Ecology and Evolution 3: 1084–1092. evolution of Andean orogeny and global climate. Earth-Science Reviews 143:1–35. Fleming TH, Geiselman C, Kress WJ. 2009. The evolution of bat pollination: a Baldwin BG, Sanderson MJ. 1998. Age and rate of diversification of the phylogenetic perspective. Annals of Botany 104: 1017–1043. Hawaiian silversword alliance (Compositae). Proceedings of the National Fulton M, Hodges SA. 1999. Floral isolation between Aquilegia formosa and Academy of Sciences, USA 95: 9402–9406. Aquilegia pubescens. Proceedings of the Royal Society of London. Series B: Biological Barthlott W, Mutke J, Rafiqpoor D, Kier G, Kreft H. 2005. Global centers of Sciences 266: 2247–2252. vascular plant diversity. Nova Acta Leopoldina 92:61–83. Garzione CN, Auerbach DJ, Jin-Sook Smith J, Rosario JJ, Passey BH, Jordan Beaulieu JM, O’Meara BC. 2015. Extinction can be estimated from moderately TE, Eiler JM. 2014. Clumped isotope evidence for diachronous surface sized molecular phylogenies. Evolution 69: 1036–1043. cooling of the Altiplano and pulsed surface uplift of the Central Andes. Earth Beaulieu JM, O’Meara BC, Crane P, Donoghue MJ. 2015. Heterogenous rates and Planetary Science Letters 393: 173–181. of molecular evolution and diversification could explain the Triassic age GarzioneCN,HokeGD,LibarkinJC,WithersS,MacFaddenB,EilerJ,GhoshP, estimate for angiosperms. Systematic Biology 64: 869–878. MulchA.2008.RiseoftheAndes.Science(NewYork,N.Y.)320:1304–1307. Bell CD, Soltis DE, Soltis PS. 2010. The age and diversification of the Garzione CN, Molnar P, Libarkin JC, MacFadden BJ. 2006. Rapid late angiosperms re-revisited. American Journal of Botany 97: 1296–1303. Miocene rise of the Bolivian Altiplano: evidence for removal of mantle Blois JL, Zarnetske PL, Fitzpatrick MC, Finnegan S. 2013. Climate change and lithosphere. Earth and Planetary Science Letters 241: 543–556. the past, present, and future of biotic interactions. Science 341: 499–504. Gentry AH. 1982. Neotropical floristic diversity: phytogeographical connections Bouchenak-Khelladi Y, Onstein RE, Xing Y, Schwery O, Linder HP. 2015. On between Central and South America, Pleistocene climatic fluctuations, or an the complexity of triggering evolutionary radiations. New Phytologist 207: accident of the Andean orogeny? Annals of the Missouri Botanical Garden 69: 313–326. 557–593. Bouckaert R, Heled J, Kuhnert€ D, Vaughan T, Wu CH, Xie D, Suchard MA, Givnish TJ. 2010. Ecology of plant speciation. Taxon 59: 1326–1366. Rambaut A, Drummond AJ. 2014. BEAST 2: a software platform for Bayesian Givnish TJ, Barfuss MHJ, Van Ee B, Riina R, Schulte K, Horres R, Gonsiska evolutionary analysis. PLoS Computational Biology 10:1–6. PA, Jabaily RS, Crayn DM, Smith JAC et al. 2014. Adaptive radiation, Cellinese N, Smith SA, Edwards EJ, Kim S-T, Haberle RC, Avramakis M, correlated and contingent evolution, and net species diversification in Donoghue MJ. 2009. Historical biogeography of the endemic Campanulaceae Bromeliaceae. Molecular Phylogenetics and Evolution 71:55–78. of Crete. Journal of Biogeography 36: 1253–1269. Givnish TJ, Millam KC, Mast AR, Paterson TB, Theim TJ, Hipp AL, Henss Clague D. 1996. Growth and subsidence of the Hawaiian-Emperor volcanic JM, Smith JF, Wood KR, Sytsma KJ. 2009. Origin, adaptive radiation and chain. In: Keast A, Mille S, eds. The origin and evolution of Pacific island biotas, diversification of the Hawaiian lobeliads (Asterales: Campanulaceae). New Guinea to Eastern Polynesia: patterns and processes. Amsterdam, the Proceedings of the Royal Society of London. Series B: Biological Sciences 276: 407– Netherlands: SPB Academic Press, 35–50. 416. Colwell RK, Betts BJ, Bunnell P, Carpenter FL, Feinsinger P. 1974. Givnish TJ, Pires JC, Graham SW, McPherson MA, Prince LM, Patterson Competition for the nectar of Centropogon valerii by the hummingbird Colibri TB, Rai HS, Roalson EH, Evans TM, Hahn WJ et al. 2005. Repeated thalassinus and the flower-piercer Diglossa plumbea, and its evolutionary evolution of net venation and fleshy fruits among monocots in shaded implications. The Condor 76: 447–452. habitats confirms a priori predictions: evidence from an ndhF phylogeny. Condamine FL, Rolland J, Morlon H. 2013. Macroevolutionary perspectives to Proceedings of the Royal Society of London. Series B: Biological Sciences 272: environmental change. Ecology Letters 16:72–85. 1481–1490. Crowl AA, Mavrodiev E, Mansion G, Haberle R, Pistarino A, Kamari G, Phitos Gregory-Wodzicki KM. 2000. Uplift history of the Central and Northern Andes: D, Borsch T, Cellinese N. 2014. Phylogeny of Campanuloideae a review. Geological Society of America Bulletin 112: 1091–1105. (Campanulaceae) with emphasis on the utility of nuclear pentatricopeptide Guzm an B, Lledo MD, Vargas P. 2009. Adaptive radiation in Mediterranean repeat (PPR) genes. PLoS ONE 9: e94199. Cistus (Cistaceae). PLoS ONE 4: e6362. Davis MP, Midford PE, Maddison W. 2013. Exploring power and parameter von Hagen KB, Kadereit JW. 2001. The phylogeny of Gentianella estimation of the BiSSE method for analyzing species diversification. BMC (Gentianaceae) and its colonization of the southern hemisphere as revealed by Evolutionary Biology 13: 38. nuclear and chloroplast DNA sequence variation. Organisms Diversity & Donoghue MJ, Sanderson MJ. 2015. Confluence, synnovation, and depauperons Evolution 1:61–79. in plant diversification. New Phytologist 207: 260–274. van der Hammen T, Cleef AM. 1986. Development of the high Andean p aramo Drummond CS, Eastwood RJ, Miotto STS, Hughes CE. 2012. Multiple flora and vegetation. In: Vuillemier F, Monasterio M, eds. High altitude tropical continental radiations and correlates of diversification in Lupinus biogeography. New York, NY, USA: Oxford University Press, 153–201. (Leguminosae): testing for key innovation with incomplete taxon sampling. Hoorn C, Mosbrugger V, Mulch A, Antonelli A. 2013. Biodiversity from Systematic Biology 61: 443–460. mountain building. Nature Geoscience 6: 154. Ehlers TA, Poulsen CJ. 2009. Influence of Andean uplift on climate and Hoorn C, Wesselingh FP, ter Steege H, Bermudez MA, Mora A, Sevink J, paleoaltimetry estimates. Earth and Planetary Science Letters 281: 238–248. Sanmart ın I, Sanchez-Meseguer A, Anderson CL, Figueiredo JP et al. 2010.
New Phytologist (2016) 210: 1430–1442 Ó 2016 The Authors www.newphytologist.com New Phytologist Ó 2016 New Phytologist Trust New Phytologist Research 1441
Amazonia through time: andean uplift, climate change, landscape evolution, Muchhala N. 2003. Exploring the boundary between pollination syndromes: bats and biodiversity. Science (New York, N.Y.) 330: 927–931. and hummingbirds as pollinators of Burmeistera cyclostigmata and B. tenuiflora Hughes C, Eastwood R. 2006. Island radiation on a continental scale: (Campanulaceae). Oecologia 134: 373–380. exceptional rates of plant diversification after uplift of the Andes. Proceedings of Muchhala N. 2006. The pollination biology of Burmeistera (Campanulaceae): the National Academy of Sciences, USA 103: 10334–10339. specialization and syndromes. American Journal of Botany 93: 1081–1089. Hughes CE, Atchison GW. 2015. The ubiquity of alpine plant radiations: from Muchhala N, Potts MD. 2007. Character displacement among bat-pollinated the Andes to the Hengduan Mountains. New Phytologist 207: 275–282. flowers of the genus Burmeistera: analysis of mechanism, process and pattern. Hughes CE, Nyffeler R, Linder HP. 2015. Evolutionary plant radiations: where, Proceedings of the Royal Society of London. Series B: Biological Sciences 274: when, why and how? New Phytologist 207: 249–253. 2731–2737. Kay KM, Reeves PA, Olmstead RG, Schemske DW. 2005. Rapid speciation and Mulch A. 2016. Stable isotope paleoaltimetry and the evolution of landscapes and the evolution of hummingbird pollination in Neotropical Costus subgenus life. Earth and Planetary Science Letters 433: 180–191. Costus (Costaceae): evidence from nrDNA ITS and ETS sequences. American Mulch A, Uba CE, Strecker MR, Schoenberg R, Chamberlain CP. 2010. Late Journal of Botany 92: 1899–1910. Miocene climate variability and surface elevation in the central Andes. Earth Klak C, Khunou A, Reeves G, Hedderson T. 2003. A phylogenetic hypothesis and Planetary Science Letters 290: 173–182. for the Aizoaceae (Caryophyllales) based on four plastid DNA regions. Myers N, Mittermeier RA, Mittermeier CG, Fonseca GAB, Kent J. 2000. American Journal of Botany 90: 1433–1445. Biodiversity hotspots for conservation priorities. Nature 403: 853–858. Kozak KH, Weisrock DW, Larson A. 2006. Rapid lineage accumulation in a Nee S, May RM, Harvey PH. 1994. The reconstructed evolutionary process. non-adaptive radiation: phylogenetic analysis of diversification rates in eastern Philosophical transactions of the Royal Society of London Series B: Biological North American woodland salamanders (Plethodontidae: Plethodon). Sciences 344: 305–311. Proceedings of the Royal Society of London. Series B: Biological Sciences 273: 539– Ng J, Smith SD. 2014. How traits shape trees: new approaches for detecting 546. character state-dependent lineage diversification. Journal of Evolutionary Biology Kreft H, Jetz W. 2007. Global patterns and determinants of vascular 27: 2035–2045. plant diversity. Proceedings of the National Academy of Sciences, USA 104: Pagel M, Meade A. 2006. Bayesian analysis of correlated evolution of discrete 5925–5930. characters by reversible-jump Markov chain Monte Carlo. The American Lagomarsino LP, Antonelli A, Muchhala N, Timmermann A, Mathews S, Davis Naturalist 167: 808–825. CC. 2014. Phylogeny, classification, and fruit evolution of the species-rich Parra M, Mora A, Jaramillo C, Strecker MR, Sobel ER, Quiroz L, Rueda M, Neotropical bellflowers (Campanulaceae: Lobelioideae). American Journal of Torres V. 2009. Orogenic wedge advance in the northern Andes: evidence Botany 101: 2097–2112. from the Oligocene-Miocene sedimentary record of the Medina Basin, Eastern Łancucka-Srodoniowa M. 1977. New herbs described from the Tertiary of Cordillera, Colombia. Geological Society of America Bulletin 121: 780–800. Poland. Acta Palaeobotanica 18:37–44. Pennington RT, Lavin M, Sarkinen T, Lewis GP, Klitgaard BB, Hughes CE. Lapoint RT, Magnacca KN, O’Grady PM. 2014. Phylogenetics of the 2010. Contrasting plant diversification histories within the Andean biodiversity Antopocerus-Modified Tarsus clade of Hawaiian Drosophila: diversification hotspot. Proceedings of the National Academy of Sciences, USA 107: 13783– across the Hawaiian Islands. PLoS ONE 9: e113227. 13787. Leier A, McQuarrie N, Garzione C, Eiler J. 2013. Stable isotope evidence for Quental TB, Marshall CR. 2010. Diversity dynamics: molecular phylogenies multiple pulses of rapid surface uplift in the Central Andes, Bolivia. Earth and need the fossil record. Trends in Ecology & Evolution 25: 434–441. Planetary Science Letters 371–372:49–58. Rabosky DL. 2010. Extinction rates should not be estimated from molecular Maddison WP, FitzJohn RG. 2015. The unsolved challenge to phylogenies. Evolution; International Journal of Organic Evolution 64: 1816– phylogenetic correlation tests for categorical characters. Systematic Biology 64: 1824. 127–136. Rabosky DL. 2014. Automatic detection of key innovations, rate shifts, and Maddison WP, Midford PE, Otto SP. 2007. Estimating a binary character’s diversity-dependence on phylogenetic trees. PLoS ONE 9: e89543. effect on speciation and extinction. Systematic Biology 56: 701–710. Rabosky DL. 2015. Challenges in the estimation of extinction from molecular Madrin~ an S, Cort es AJ, Richardson JE. 2013. P aramo is the world’s fastest phylogenies: a response to Beaulieu and O’Meara. Evolution 70: 218–228. evolving and coolest biodiversity hotspot. Frontiers in Genetics 4:1–7. Rabosky DL, Donnellan SC, Talaba AL, Lovette IJ. 2007. Exceptional among- Magall on S, Sanderson MJ. 2001. Absolute diversification rates in angiosperm lineage variation in diversification rates during the radiation of Australia’s most clades. Evolution 55: 1762–1780. diverse vertebrate clade. Proceedings of the Royal Society of London. Series B: McGuire JA, Witt CC, Altshuler DL, Remsen JV. 2007. Phylogenetic Biological Sciences 274: 2915–2923. systematics and biogeography of hummingbirds: Bayesian and maximum Rabosky DL, Goldberg EE. 2015. Model inadequacy and mistaken inferences of likelihood analyses of partitioned data and selection of an appropriate trait-dependent speciation. Systematic Biology 64: 340–355. partitioning strategy. Systematic Biology 56: 837–856. Rabosky DL, Grundler M, Anderson C, Title P, Shi JJ, Brown JW, Huang H, McGuire JA, Witt CC, Remsen JV, Corl A, Rabosky DL, Altshuler DL, Dudley Larson JG. 2014. BAMMtools: an R package for the analysis of evolutionary R. 2014. Molecular phylogenetics and the diversification of hummingbirds. dynamics on phylogenetic trees (Kembel S, ed.). Methods in Ecology and Current Biology 24: 910–916. Evolution 5: 701–707. Merckx VSFT, Hendriks KP, Beentjes KK, Mennes CB, Becking LE, Rambaut A, Suchard MA, Xie D, Drummond AJ. 2014. Tracer v1.6. [WWW Peijnenburg KTC, Afendy A, Arumugam N, de Boer H, Biun A et al. document] URL http://beast.bio.ed.ac.uk/Tracer. 2015. Evolution of endemism on a young tropical mountain. Nature 524: Richardson JE, Pennington RT, Pennington TD, Hollingsworth PM. 2001. 347–350. Rapid diversification of a species-rich genus of neotropical rain forest trees. Moore W, Robertson JA. 2014. Explosive adaptive radiation and extreme Science (New York, N.Y.) 293: 2242–2245. phenotypic diversity within ant-nest beetles. Current Biology 24:1–5. Salzburger W, Mack T, Verheyen E, Meyer A. 2005. Out of Tanganyika: Morlon H, Lewitus E, Condamine FL, Manceau M, Clavel J, Drury J. 2016. genesis, explosive speciation, key-innovations and phylogeography of the RPANDA: an R package for macroevolutionary analyses on phylogenetic trees. haplochromine cichlid fishes. BMC Evolutionary Biology 5: 17. Methods in Ecology and Evolution, doi: 10.1111/2041-210X.12526. S€arkinen T, Pennington RT, Lavin M, Simon MF, Hughes CE. 2012. Morlon H, Parsons TL, Plotkin JB. 2011. Reconciling molecular phylogenies Evolutionary islands in the Andes: persistence and isolation explain high with the fossil record. Proceedings of the National Academy of Sciences, USA 108: endemism in Andean dry tropical forests. Journal of Biogeography 39: 16327–16332. 884–900. Moyle RG, Filardi CE, Smith CE, Diamond J. 2009. Explosive Pleistocene Scherson RA, Vidal R, Sanderson MJ. 2008. Phylogeny, biogeography, and rates diversification and hemispheric expansion of a ‘great speciator’. Proceedings of of diversification of New World Astragalus (Leguminosae) with an emphasis on the National Academy of Sciences, USA 106: 1863–1868. South American radiations. American Journal of Botany 95: 1030–1039.
Ó 2016 The Authors New Phytologist (2016) 210: 1430–1442 New Phytologist Ó 2016 New Phytologist Trust www.newphytologist.com New 1442 Research Phytologist
Schiestl FP, Schluter€ PM. 2009. Floral isolation, specialized pollination, and Fig. S1 Time-calibrated phylogeny of Neotropical Campanu- – pollinator behavior in orchids. Annual Review of Entomology 54: 425 446. laceae, with outgroup representatives of Lobelioideae and Cam- Schmidt-Lebuhn AN, Kessler M, Hensen I. 2007. Hummingbirds as drivers of panuloideae collapsed. plant speciation? Trends in Plant Science 12: 329–331. Schwery O, Onstein RE, Bouchenak-Khelladi Y, Xing Y, Carter RJ, Linder HP. 2015. As old as the mountains: the radiations of the Ericaceae. New Phytologist Fig. S2 Time-calibrated phylogeny of Neotropical Campanu- 207: 355–367. laceae, with the ingroup collapsed and showing outgroup rela- Sempere T, Hartley A, Roperch P. 2006. Comment on ‘Rapid uplift of the tionships. Altiplano revealed through 13C-18O bonds in paleosol carbonates’. Science (New York, N.Y.) 314: 760. Smith JF. 2001. High species diversity in fleshy-fruited tropical understory Fig. S3 Phylogram showing molecular-proportional branch plants. The American Naturalist 157: 646–653. lengths, illustrating substitution rate heterogeneity within clade. Smith SA, Donoghue MJ. 2008. Rates of molecular evolution are linked to life – history in flowering plants. Science (New York, N.Y.) 322:86 89. Fig. S4 Diversification rate analysis for Neotropical Campanu- Smith SA, O’Meara BC. 2012. treePL: divergence time estimation using laceae in BAMM. penalized likelihood for large phylogenies. Bioinformatics 28: 2689–2690. Spriggs EL, Clement WL, Sweeney PW, Madrin~ an S, Edwards EJ, Donoghue MJ. 2015. Temperate radiations and dying embers of a tropical past: the Fig. S5 Posterior distributions of trait-dependent speciation, diversification of Viburnum. New Phytologist 207: 340–354. extinction, transition, and net diversification rates as estimated by Stamatakis A. 2014. RAxML version 8: a tool for phylogenetic analysis and post- BiSSE. analysis of large phylogenies. Bioinformatics 30: 1312–1313. Stein BA. 1992. Sicklebill hummingbirds, ants, and flowers: plant–animal interactions and evolutionary relationships in Andean Lobeliaceae. BioScience Table S1 Taxon sampling including voucher information, coun- 42:27–33. try of origin, and GenBank accession numbers for seven plastid Theim TJ, Shirk RY, Givnish TJ. 2014. Spatial genetic structure in four loci understory Psychotria species (Rubiaceae) and implications for tropical forest – diversity. American Journal of Botany 101: 1189 1199. Table S2 Divergence time estimation to assess sensitivity to dat- Thiers B. 2013. Index Herbariorum: a global directory of public herbaria and associated staff. New York Botanical Garden’s Virtual Herbarium. [WWW ing method, calibration selection, and branch length heterogeni- document] URL http://sweetgum.nybg.org/ih (continuously updated). ety Valente LM, Etienne RS, Phillimore AB. 2014. The effects of island ontogeny on species diversity and phylogeny. Proceedings of the Royal Society of London. Table S3 Trait codings for the four traits used in our BiSSE anal- Series B: Biological Sciences 281: 20133227. yses Valente LM, Savolainen V, Vargas P. 2010. Unparalleled rates of species diversification in Europe. Proceedings of the Royal Society of London. Series B: Biological Sciences 277: 1489–1496. Table S4 Paleoenvironmental-dependent diversification analyses Wagner CE, Harmon LJ, Seehausen O. 2012. Ecological opportunity and sexual using paleoaltimetry and Cenozoic climate data selection together predict adaptive radiation. Nature 487: 366–369. Walker JD, Geissman JW, Bowring SA, Babcock LE. 2012. Geologic Time Scale Table S5 Model comparison for the four BiSSE analyses pre- v. 4.0. Geological Society of America, doi: 10.1130/2012.CTS004R3C. Weber MG, Agrawal AA. 2014. Defense mutualisms enhance plant sented in the main text diversification. Proceedings of the National Academy of Sciences, USA 111: – 16442 16447. Table S6 Correlated evolution of binary traits using BAYESTRAITS Wimmer FE. 1943. Campanulaceae-Lobelioideae. I. Teil. In: Mansfeld R, ed. – Das Pflanzenreich IV.276b. Leipzig, Germany: Wilhem Engelmann, 1 260. Table S7 Model comparison for the two diversity-dependence Wimmer FE. 1953. Campanulaceae-Lobelioideae, II. Teil. In: Mansfeld R, ed. Das Pflanzenreich IV.276b. Berlin, Germany: Akademie-Verlag, 261–814. analyses presented, with mean parameter estimates for each Zachos JC, Dickens GR, Zeebe RE. 2008. An early Cenozoic perspective on model greenhouse warming and carbon-cycle dynamics. Nature 451: 279–283. Please note: Wiley Blackwell are not responsible for the content Supporting Information or functionality of any supporting information supplied by the authors. Any queries (other than missing material) should be Additional supporting information may be found in the online directed to the New Phytologist Central Office. version of this article.
See also the Commentary on this article by Hughes, 210: 1152–1154.
New Phytologist (2016) 210: 1430–1442 Ó 2016 The Authors www.newphytologist.com New Phytologist Ó 2016 New Phytologist Trust